By Evan Vega
NetworkChuck’s sponsored video ran DeepSeek V4.1 Flash, about 748 billion parameters, on WiCi One’s 32 GB RTX 5090 at roughly eight tokens a second by streaming the model from an SSD. Novel Cognition checked 21 of its claims, 12 hold and none fail, then tested a 256 GB Mac Studio: the smaller DeepSeek V4 Flash in memory at 22.5 tokens a second, a model switch in 0.69 seconds and a streamed answer using 18 kilobits a second.
Related: Frontier Watch